Testing War In the Error Term

Abstract

The proof for “War Is in the Error Term,” a piece that appeared in the Summer 1999 issue of International Organization, contains a subtle error. Once the correction is made, there are broader implications for testing theories of war using quantitative studies. Large-n tests to verify the incomplete information explanation for war will be more difficult to perform than originally anticipated.

Global and Diplomatic Political Risks and Foreign Direct Investment

“…Since both host country's cost of signaling and home country's cost of fighting increase with the economic benefits derived from the presence of this FDI, it is more likely that both countries will prefer a negotiated settlement to escalation as their economic interdependence grows. The host country, by not resorting to sanctions, will reveal its lack of resolve and the home country, the initiator of the dispute, will make a more generous offer as it takes into account its own war costs, both behaviors expanding the probability that the home country will make an offer equal to or above the minimum offer that the host country will accept (Boehmer et al, 2001; Coletta and Gartzke, 2003; Fearon, 1995). 20 Assuming that these benefits positively depend on the value of U.S. assets in the host country and that the marginal productivity of foreign capital is decreasing, 21 a low ROA induced by a large stock of U.S. FDI may be thus negatively correlated with the risk of diplomatic dispute.…”

Interdependence and the Duration of Militarized Conflict

“…The inverse relationship between the costs and probability of conflict can also be derived formally from strategic interaction models, although the result is not insensitive to the model specification Reed (2003). analyzes an ultimatum model with one-sided uncertainty over the distribution of power, in which an increase in any state's conflict costs leads to a reduced probability of disagreement Coletta & Gartzke (2003). analyze an ultimatum model with one-sided uncertainty over the costs of the demand respondent and find that, while the probability of conflict is always decreasing in the costs of the demand sender, the effect of the respondent's costs is contingent upon the assumptions one makes about the belief distribution function.…”